Due in 12 hours

The effect of corporate social responsibility on stock performance: new evidence for the USA and

Europe

URS VON ARXyz and ANDREAS ZIEGLER*yx yCenter for Corporate Responsibility and Sustainability, University of Zurich, Zähringerstr. 24, 8001 Zurich, Switzerland zCenter of Economic Research, Swiss Federal Institute of Technology (ETH) Zurich, Zürichbergstr. 18, 8032 Zurich,

Switzerland xDepartment of Economics, University of Kassel, Nora-Platiel-Str. 5, 34109 Kassel, Germany

(Received 22 June 2010; in final form 11 June 2013)

This paper provides new empirical evidence for the effect of corporate social responsibility on corporate financial performance. In contrast to former studies, we examine two different regions, namely the USA and Europe, and disentangle firm and sector specific impacts. Our econometric analysis shows that environmental and social activities of a firm compared with other firms within the industry are valued by financial markets in both regions. However, the respective posi- tive effects on average monthly stock returns between 2003 and 2006 are more robust in the USA and, in addition, non-linear. Our analysis furthermore points to biased parameter estimates if incorrectly specified econometric models are applied: the seemingly significantly negative effect of environmental and social performance of the industry to which a firm belongs strongly declines and mostly becomes insignificant if the explanation of stock performance is based on the Fama–French three-factor or the Carhart four-factor models instead of the simple Capital Asset Pricing Model.

Keywords: Corporate social responsibility; Financial performance; Asset pricing models

JEL Classification: G12, M14, Q01, Q56

1. Introduction

This paper empirically studies economic effects of ‘corporate social performance’ or ‘corporate social responsi- bility’ (CSR). These terms are typically used synonymously and comprise both corporate social as well as corporate environmental activities (e.g. Waddock and Graves 1997, Orlitzky 2001, Orlitzky et al. 2003). According to the popu- lar definition of McWilliams and Siegel (2001), CSR is described as the ‘actions that appear to further some social good, beyond the interest of the firm and what is required by law’. Another definition of CSR refers to ‘actions which reduce the extent of externalized costs’ (besides the avoid- ance of distributional conflicts, Heal 2005). In line with some former studies, we consider the financial impacts of CSR, i.e. the effects of CSR on stock performance which is used as an indicator for corporate financial performance.

Our main contribution to the corresponding literature is twofold: first of all, we examine this relationship in two dif- ferent regions, namely the USA and Europe. In this respect, we are particularly able to incorporate the same CSR indi- cators for both regions. This allows a comparative analysis for these two world-wide leading stock markets. Therefore, we can analyse whether potential CSR impacts are interregional or whether regional differences arise. Sec- ondly, we apply different asset pricing models for the expla- nation of stock performance, i.e. the three-factor model according to Fama and French (1993) and particularly the four-factor model according to Carhart (1997) beside the simple Capital Asset Pricing Model (CAPM). While the corresponding factors for these models are publicly avail- able for the USA and some other specific stock markets, they have to be calculated for the entire European stock market. This is obviously the reason why such multifactor models have not often been applied for this region yet.

Knowledge about the effect of corporate environmental performance as one component of CSR on corporate*Corresponding author. Email: [email protected]

� 2013 Taylor & Francis

Quantitative Finance, 2014

Vol. 14, No. 6, 977–991, http://dx.doi.org/10.1080/14697688.2013.815796

financial performance contributes to the debate about whether managers systematically miss profit opportunities if they decide against the protection of the natural environ- ment (e.g. King and Lenox 2002). This debate has been going on for a while in the corresponding literature (e.g. Hart and Ahuja 1996, King and Lenox 2001, Guenster et al. 2011). Furthermore, an understanding of this relation- ship is also interesting for environmental policy: if a posi- tive effect of corporate environmental performance really exists, it can be argued that traditional mandatory command and control regulations as well as market-based instru- ments—such as green taxes—should be relaxed. Instead, these regulations could be supplemented or even substituted by information-based instruments, namely by improving the flow of information with respect to this effect (e.g. Telle 2006). Just like other non-mandatory approaches in environmental policy—such as voluntary green management measures—these regulations can be thought to be more cost-efficient because they leave firms the flexibility to choose the cheapest pollution abatement strategy and reduce governments’ enforcement costs (e.g. Alberini and Segerson 2002). All these conclusions do not only apply for the effect of corporate environmental, but also corporate social activities and thus CSR on corporate financial performance.

Due to the inconclusiveness of theory, the effect of CSR on corporate financial performance in general and on stock performance in particular is ultimately an empirical ques- tion. Against this background, we provide new empirical evidence for this issue. As an indicator for CSR, we use environmental and social activities of a firm compared with other firms in the same industry. In line with only few former studies (e.g. King and Lenox 2001, Ziegler et al. 2007a), we additionally consider sector-specific influences by incorporating environmental and social performance of the industry to which a firm belongs. As an indicator for stock performance, we consider the average monthly stock returns between 2003 and 2006. Due to this specific depen- dent variable, our final cross-sectional regressions have to be based on the estimation of asset pricing models since financial economics suggests the use of corresponding fac- tors to explain average stock returns.

The remainder of the paper is structured as follows: Section 2 briefly provides some theoretical background. In the third section, we review the empirical literature regard- ing the relationship between CSR and corporate financial performance. Section 4 discusses our different two-stage econometric approaches. In the fifth section, the used data and variables are described. Section 6 reports the empirical results and the final section discusses our results and concludes.

2. Theoretical background

Overall, current theory concerning the effects of CSR on corporate financial performance is ambiguous (e.g. Waddock and Graves 1997, Guenster et al. 2011). Argu- ments for a negative influence can be based on neoclassical microeconomics. According to this, it is mainly emphasized that the operating costs of corporate environmental (e.g.

Telle 2006) or social activities outweigh their financial benefits due to cost reductions. Therefore, CSR can lead to reduced profits, decreased firm values or competitive disad- vantage besides lower shareholder returns. This neoclassical argumentation is supported by corporate governance theory (e.g. Shleifer and Vishny 1997, Tirole 2006). According to a rather narrow definition, corporate governance comprises all measures which assure that investors get an adequate return for their investments. According to this, it is argued that, for example, the consideration of goals of other groups—such as the general public—as motivation for corporate environmental and social activities unnecessarily enlarges the latitude of management which is misused for maximizing the utility of managers. Therefore, investors have to reckon with lower returns on their investments if the respective corporations deviate from the optimal path due to excessive environmental or social activities (e.g. Heinkel et al. 2001, Beltratti 2005).

However, positive effects of CSR on corporate financial performance can also be based on neoclassical microeco- nomics by emphasizing the role of CSR in reducing the extent of externalized costs. While, for example, Friedman (1970) assumes that the government defines property rights such that no external effects exist, Heal (2005) argues that the government does not fully resolve all problems with external effects and that the competitive markets are not efficient. Therefore, CSR can substitute missing markets if external effects arise from them and can reduce conflicts between firms and stakeholder groups such as the govern- ment, non-governmental organizations, employees or cli- ents. As a consequence, it can be argued that the reduction of these conflicts increases corporate profits or corporate financial performance at least in the long run which also makes firms with a high intensity of CSR more attractive to investors.

This stakeholder argument is strengthened in the strategic management literature (e.g. Waddock and Graves 1997, Barnett and Salomon 2006, Curran and Moran 2007). Stakeholder theory suggests that management has to satisfy several groups who have some interest or ‘stake’ in a firm and can influence its outcome (e.g. McWilliams et al. 2006). Regarding corporate financial performance, it can, therefore, be worthwhile to engage in CSR because other- wise these stakeholders could withdraw the support for the firm. For example, a minimum of corporate environmental activities and the avoidance of child labour in the full value-added chain of the products can reduce risk due to, for example, aggressive campaigns of non-governmental organizations. These arguments from stakeholder theory can be embedded in the resource-based view of the firm (e.g. Barney 1991) which suggests that competitive advantage evolves from internal capabilities which are valuable, rare and difficult to imitate or substitute (e.g. Russo and Fouts 1997, Klassen and Whybark 1999, King and Lenox 2001, McWilliams et al. 2006). In this respect, stakeholder man- agement can be considered an important organizational capability or resource.

This discussion of possible positive and negative effects of CSR also implies that they can be different across sev- eral countries if some aspects play a more important role in

978 U. von Arx and A. Ziegler

specific regions. It can, for example, be argued that the stakeholder theory argument is of high relevance for Europe since the higher awareness of climate change and stringency of climate policy in this region (at least in the observation period until 2006) can lead to a high institutional pressure (e.g. by non-governmental organizations) that demands cor- porate responses to climate change (e.g. by reducing corpo- rate greenhouse gas emissions, e.g. Ziegler et al. 2011). This would lead to a more positive effect of such corporate responses as component of CSR in Europe compared with the USA. On the other hand, however, it can also be argued that the USA has a much longer tradition of ethical compo- nents of CSR and particularly socially responsible investing (SRI) than Europe. For example, (ethically and religiously motivated) SRI in the USA has already its roots in the eigh- teenth century. This long tradition would rather support a more positive effect of CSR on financial performance in the USA.

Based on this discussion of positive and negative effects of CSR, it can also be argued that there are many converse corporate environmental and social factors such that no sig- nificant effect exists (e.g. Waddock and Graves 1997, Elsayed and Paton 2005). As a consequence, the effect of CSR on corporate financial performance is ultimately an empirical question. Regarding the measurement of corporate financial performance, the use of forward-looking average stock returns is theoretically attractive in this respect. According to the efficient market hypothesis, stock prices reflect all publicly available information about the future financial performance of firms. In compliance with the well-known dividend discount model, a stock price, there- fore, equals the discounted expected future stream of divi- dends paid to the shareholders. Our approach, therefore, has the advantage that the focus is not on the past realized but on the future expected financial performance (e.g. Miller and Modigliani 1961).

3. Empirical literature review

The relationship between CSR and corporate financial per- formance can be empirically analysed with three methodo- logical approaches, namely portfolio analyses, event studies and longer term (micro-) econometric approaches. Portfolio analyses in this field typically compare the risk-adjusted stock returns of portfolios that consist of corporations with a higher environmental or social performance and portfolios that consist of stock corporations with a lower environmen- tal or social performance. Recent studies are mostly based on the estimation of alphas within multifactor models such as the Carhart four-factor model (e.g. Derwall et al. 2005, Bauer et al. 2005, 2007, Kempf and Osthoff 2007, Ziegler et al. 2011). However, the identification of isolated causal effects of CSR on corporate financial performance needs more sophisticated approaches. In this respect, event studies examine the mean stock returns for corporations experienc- ing a specific event (i.e. new information) and, therefore, aim to measure the effect on the value of a corporation (e.g. MacKinlay 1997, Kothari and Warner 2006). In the

meantime, a growing number of CSR-related event studies is available (e.g. Hamilton 1995, Posnikoff 1997, Dasgupta et al. 2001, Gupta and Goldar 2005, Curran and Moran 2007). However, one main weakness of event studies is that the application of them in general depends on unexpected events.

But, CSR rather refers to long-term corporate activities and thus cannot often only be characterized by unex- pected positive or negative events. As a consequence, longer term econometric approaches have received increasing attention for a while. These studies apply—in the same way as portfolio analyses and event studies— very different indicators for CSR. They additionally use different indicators for corporate financial performance. Due to the theoretical attractiveness as discussed above, some of these studies (e.g. Filbeck and Gorman 2004, Ziegler et al. 2007a) also use stock returns as they are a constitutive element in portfolio analyses and event stud- ies. In contrast, most other studies (e.g. Hart and Ahuja 1996, Russo and Fouts 1997, Waddock and Graves 1997, McWilliams and Siegel 2000, King and Lenox 2001, 2002, Elsayed and Paton 2005, Telle 2006, Becchetti et al. 2008, Guenster et al. 2011, Ziegler 2012) also apply accounting data-based indicators for corporate finan- cial performance such as Tobin’s Q, return on assets, return on sales or return on equity.

Using a broad measurement of CSR, we consider both corporate environmental and social activities. In contrast, many former studies neglect the social dimension of CSR by using one-dimensional indicators such as emissions of pollutants (e.g. Hart and Ahuja 1996, King and Lenox 2001, 2002, Telle 2006). However, such emission data seem to be a weak indicator for CSR in general because they only give information about a single constituent of corporate environmental performance. Other econometric analyses use more general CSR indicators which, how- ever, only refer to the environmental dimension (e.g. Rus- so and Fouts 1997, Filbeck and Gorman 2004, Elsayed and Paton 2005, Guenster et al. 2011). Studies which also incorporate CSR dimensions such as Waddock and Graves (1997), McWilliams and Siegel (2000), Ziegler et al. (2007a) and Becchetti et al. (2008) are exceptions in this respect. Furthermore, our study disentangles—such as King and Lenox (2001) and Ziegler et al. (2007a)— firm- and sector-specific influences and, therefore, addi- tionally analyses industry environmental and social performance.

4. Econometric approach

Our study applies cross-sectional regressions of average stock returns over time on CSR. To explain such stock performance, we include some control variables which are received by time-series regressions of asset pricing models. The estimated corporate beta parameters from this first stage are then—in addition to the mainly interesting CSR variables—incorporated in the final cross-sectional regressions.

The effect of corporate social responsibility 979

4.1. Time-series regressions of asset pricing models

So far, the main asset pricing model for estimating stock returns is the one-factor model based on the market model (e.g. Sharpe 1963) and the CAPM (e.g. Lintner 1965, Fama and French 2004, Perold 2004). This model can be formu- lated for a corporation or stock i in month t (i ¼ 1; . . . ; N; t ¼ 1; . . . ; T) as follows:

reit ¼ aCAPMi þ bCAPMi remt þ eit (1)

The excess returns are reit ¼ rit � rft and remt ¼ rmt � rft. In this approach, rit and rmt are the (continuous) stock returns for corporation i and the market at the end of month t, rft is the risk-free interest rate at the beginning of month t, and ɛit is the disturbance term with expectation zero and (unknown) variance r2e . Finally, a

CAPM i and b

CAPM i are fur-

ther unknown parameters which are estimated by ordinary least squares (OLS). The idea is that the estimated market-

beta parameters b̂ CAPM

i capture the non-diversifiable risk of each stock which can be used in the final cross-sectional regressions to explain average stock returns over time.

Based on the ‘anomalies’ discussion which questions the validity of the CAPM (e.g. Banz 1981, DeBondt and Thaler 1985, Fama and French 1992), Fama and French (1993) developed a three-factor model which includes—besides the excess returns rmt�rft of the market—two additional factors with respect to size and value to explain the excess returns rit�rft:

reit ¼ aFFi þ bFFi1 remt þ bFFi2 SMBt þ bFFi3 HMLt þ eit (2)

In this model, SMBt is (at the end of month t) the differ- ence between the returns for portfolios comprising stocks of ‘small’ corporations and portfolios comprising stocks of ‘big’ corporations. HMLt is (at the end of month t) the dif- ference between the returns for portfolios comprising stocks of corporations with a ‘high’ book-to-market equity and portfolios comprising stocks of corporations with a ‘low’ book-to-market equity. The main unknown parameters are now aFFi ; b

FF i1 ; b

FF i2 ; and b

FF i3 . Many studies show that this

three-factor model has more explanatory power than the one-factor model based on the CAPM, for example, Fama and French (1993, 1996) for the USA, Berkowitz and Qiu (2001) for the Canadian, Hussain et al. (2002) for the British and Ziegler et al. (2007b) for the German stock market.

Almost at the same time, however, a broad discus- sion about another factor, namely the momentum factor, began (e.g. Jagadeesh and Titman 1993, Rouwenhorst 1998, Jagadeesh and Titman 2001). As a consequence, the following four-factor model of Carhart (1997) which additionally includes this momentum factor—besides the three Fama-French factors—is in the meantime, due to the highest explanatory power, the most common asset pricing model for applications in financial economics (e. g. L’Her et al. 2004, Bollen and Busse 2005) and par- ticularly, as discussed above, for portfolio analyses on the relationship between CSR and stock performance:

reit ¼ aCARi þ bCARi1 remt þ bCARi2 SMBt þ bCARi3 HMLt þ bCARi4 MOMt þ eit (3)

In this model, MOMt is (at the end of month t) the differ- ence between the returns for portfolios comprising stocks of ‘winners’ in the past and portfolios comprising stocks of ‘losers’ in the past. The main unknown parameters are now

aCARi ; b CAR i1 ; b

CAR i2 ; b

CAR i3 ; and b

CAR i4 .

4.2. Final cross-sectional regressions

The final cross-sectional regressions with the average monthly stock returns �ri between 2003 and 2006 for corpo- ration i as dependent variables incorporate—besides the mainly interesting CSR variables (including environmental and social performance of the industry to which a firm belongs), summarized in the (column) vector CSRi—the estimated beta parameters from the time-series regressions of the several asset pricing models in the first stage as explanatory variables. In other words, these regressions

either comprise b̂ CAPM

i or b̂ FF

i1 , b̂ FF

i2 and b̂ FF

i3 or b̂ CAR

i1 , b̂ CAR

i2 ,

b̂ CAR

i3 and b̂ CAR

i4 such that the following three estimation equations arise (ζi are the respective disturbance terms):

�ri ¼ c þ d0CSRi þ kb̂ CAPM

i þ fi (4)

�ri ¼ c þ d0CSRi þ k1b̂ FF

i1 þ k2b̂ FF

i2 þ k3b̂ FF

i3 þ fi (5)

�ri ¼ c þ d0CSRi þ k1b̂ CAR

i1 þ k2b̂ CAR

i2 þ k3b̂ CAR

i3 þ k4b̂ CAR

i4 þ fi (6)

The (heteroskedasticity robust OLS) estimation of the parameters (or parameter vectors) leads to ĉ, d̂ and k̂ in

the first approach based on the CAPM, to ĉ, d̂, k̂1, k̂2 and k̂3 in the second approach based on the Fama–French

three-factor model and to ĉ, d̂, k̂1, k̂2, k̂3 and k̂4 in the third approach based on the Carhart four-factor model. As the estimated beta parameters can be theoretically consid-

ered as risk factors, the corresponding estimates k̂, k̂1, k̂2,

k̂3 and k̂4 in the final cross-sectional regressions are expected to be positive.

In this respect, it should be noted that the cross-sectional regressions for the European stock market additionally include nine country dummies as further explanatory vari- ables to control for possible regional differences regarding the average stock returns over time. The corresponding vari- ables Fini, Frai, Geri, Itai, Neti, Spai, Swe, Swii and UKi take the value one if corporation i stems from Finland, France, Germany, Italy, the Netherlands, Spain, Sweden, Switzerland and the UK, respectively. The final cross-sec- tional regressions additionally comprise corporations from Austria, Belgium, Denmark, Greece, Ireland, Norway and Portugal. Due to the small number of firms from these countries, the corresponding dummy variables are not included in the regressions, but serve as a summarized

980 U. von Arx and A. Ziegler

omitted reference category for the other country dummies. All calculations for this paper were performed with the soft- ware package STATA.

5. Data and variables

5.1. CSR data and variables

Concerning the CSR variables, we use data from the Swiss bank Sarasin & Cie in Basle. Reliably beginning in 2001/ 2002, this bank has assessed environmental and social activities for 317 corporations in the USA and 720 Euro- pean corporations quoted on different stock exchanges over time. While most of these corporations are large, some of them have a very low market capitalization. The latter firms were only assessed due to their sustainability profile (from the perspective of Sarasin). However, the problem is that such diverse firms which belong to very different sectors cannot be reliably compared regarding the effect of CSR on stock performance. Therefore, we only examine those assessed firms which were member of the Morgan Stanley Capital International (MSCI) Europe Index or the MSCI US Index at least once over the period between 1996 and 2006. This period was chosen because we had only access to financial data for these 11 years as discussed below. As a consequence, the number of corporations reduces to 212 in the USA and 419 in Europe.

Indeed, the corresponding necessary financial data and the exclusively used assessments for 2002 are only avail- able for N = 173 USA and N = 268 European corporations which are finally considered in our empirical analysis. In this respect, it should be noted that only those corpora- tions are examined whose financial data are available for all 132 months over the entire period between 1996 and 2006. The reason for this is that the number T of observations should be large for the time-series regres- sions of asset pricing models. In other words, if we had additionally incorporated corporations whose financial data are only available for a lower number of months, the corresponding estimations of the beta parameters would be less reliable. Furthermore, we incorporate lagged explanatory variables by using the 2002 assessments and the average monthly stock returns between 2003 and 2006 as dependent variables in the final cross-sectional regressions.

In its assessments, Sarasin combines environmental and social risk indicators in a two-dimensional rating and, therefore, considers both activities of a firm compared with other firms in the same industry as well as environmental and social performance of the industry to which a firm belongs. These two ratings are ultimately used to deter- mine whether a corporation is suitable for Sarasin’s sus- tainable investment funds and portfolios or not. The first rating indicates how successfully firms manage the industry-specific risks. Concerning the environmental dimension of this rating, all measures of a corporation to reduce environmental risks in the full value-added chain of the products (pre-production, production, use of products or services) are assessed. Furthermore, environmental

strategies and management systems are considered. Specifi- cally, Sarasin uses sub-criteria which are proposed by the World Business Council for Sustainable Development. These environmental sub-criteria are energy intensity, use of renewable energies, material intensity, toxicity, revalorisation, durability and service intensity.

Concerning the social dimension of the first rating, it is assessed how well a firm manages its internal and external conflict potential, i.e. requirements of different stakeholder groups. The following groups are considered as stakeholders: employees, suppliers, investors, the gen- eral public as well as—regarding the market—customers and competitors. Key elements for the assessment are the social strategy and social management systems of firms. As social sub-criteria, Sarasin considers health risks, par- ticipation, wealth creation and distribution, and knowledge creation regarding their effects on stakeholder groups. These single social and environmental sub-criteria—which are assessed on a five-stage scale, respectively—are then aggregated to the first broad rating. It should be noted that the relevance of the several environmental sub-criteria differs between sectors with respect to the value-added chain of the products, whereas the relevance of the sev- eral social sub-criteria differs between sectors with respect to the importance of the individual stakeholder groups. As a consequence, the final aggregation is based on different weightings.

The second industry-specific rating refers to the assess- ment of the environmental and social impacts and risks which are particular for this sector. In this assessment, not only the direct effects of producing the products and services, but also indirect influences along the product chain as well as lifecycle considerations are included. Regarding the environmental dimension, two main sub- criteria are considered, namely resource consumption and emissions. According to these criteria, for example, pri- mary industries such as chemicals, energy, energy suppli- ers, metal production, mining, paper and cement all belong to industries with higher environmental risks with respect to the substantial direct impact of those industries on the environment.

Regarding the social dimension, Sarasin distinguishes between internal conflict potential (e.g. downsizing or inad- equate working conditions) and external conflict potential which comprises, for example, health risks caused by prod- ucts and productions methods, concentration of economic power, corruption and ethical conflicts. While each of these single environmental and social industry-specific sub-criteria is again independently assessed on a five-stage scale, the several sub-criteria are finally—based on different weigh- tings—aggregated to the second broad rating. Both aggre- gated broad ratings are also based on a five-stage scale and, therefore—in the same way as the single sub-criteria— coded with the integers from one to five. In this respect, the number one designates the worst and the number five the best assessment.

In the following, Corpi symbolizes the corresponding ordinal variable for the environmental and social activi- ties of a firm i compared with other firms in the same industry and Indui symbolizes the ordinal variable for

The effect of corporate social responsibility 981

environmental and social performance of the industry to which a firm i belongs. Since it is not certain that these ratings are equidistant in each case, i.e. that the distance between two numbers is always identical, we also examine dummy variables based on these ordinal variables. How- ever, preliminary investigations showed that the incorpora- tion of overall eight single dummies lead to ambiguous estimation results, obviously because the effects of these variables are not linear. Therefore, we analyse two alterna- tive aggregated dummy variables for both ratings in more detail. The dummies Corp54i or Indu54i take the value one if Corpi or Indui take the values five or four, respectively. Furthermore, the dummies Corp543i or Indu543i take the value one if Corpi or Indui take the values five, four or three, respectively. As a consequence, the vectors CSRi in the final cross-sectional regressions always comprise exactly one pair of the variables Corpi and Indui, Corp54i and Indu54i or Corp543i and Indu543i. Table 1 reports the corresponding frequencies for the distribution of the values of Corpi and Indui for the N = 173 USA and the N = 268 European corporations. Here, the relative frequencies (in%) for the values of Corpi refer to the respective values of Indui in the columns.

5.2. Financial variables

As aforementioned, we had access to financial data on total return indexes (which contain both stock prices and cash flows to the investor), market values and book values (in US $, respectively) for the period between 1996 and 2006. These data stem from the Thomson Financial Data- stream database. All monthly stock returns rit (in%) for both the USA and European corporations in the empirical analysis were calculated with these total return indexes. The time-series regressions in the first stage of the econo- metric analysis additionally require the inclusion of a risk-free interest rate for the calculation of excess returns. In this respect, we used the monthly return of one-month US Treasury Bills. Furthermore, the time-series regressions additionally require the inclusion of the monthly excess stock returns remt for the market. For the USA, we directly used the corresponding data (in%) from the homepage of Kenneth R. French (http://mba.tuck.dartmouth.edu/pages/ faculty/ken.french/data_ library.html). The calculation of the monthly returns rmt of a European stock market portfolio (in%) is based on the total return indexes of the MSCI Europe (in US $).

In the same way as remt, the factors SMBt, HMLt and MOMt for the US stock market were directly extracted from the homepage of Kenneth R. French. In contrast, these factors were not publicly available for the entire European stock market and thus had to be constructed. The bases for this calculation were 917 European corpora- tions which were member of the MSCI Europe at least once over the complete period between 1996 and 2006. Regarding SMBt and HMLt, corporations were ranked each year on their market capitalization in June and indepen- dently on their ratio between the published book value for the previous year and the market value in December of the previous year. Then, the median of the market

capitalizations as well as the 30 and 70% percentiles of the book-to-market equity were calculated such that six portfolios could be constructed from these three values. In each June over time, the corporations were allocated anew to one of these six portfolios and stay there from July for the subsequent 12 months.

The construction of these portfolios only comprises those corporations with corresponding available data for June of the respective year and additionally with positive book values for the previous year. Furthermore, stock return data and market value data for the subsequent 12 months had to be available. The resulting times-series of the value-weighted returns of these six stock portfolios (between July 1997 and June 2006) were the basis for the final calculations of SMBt which is the (weighted) difference between the returns of ‘small’ corporations and ‘big’ corporations as well as HMLt which is the (weighted) difference between the returns for corporations with a ‘high’ book-to-market equity and corporations with a ‘low’ book-to-market equity (according to the procedure of Fama and French 1993).

Concerning MOMt, corporations were ranked in each month t�1 on their market capitalization and independently on their average stock returns between the months t�12 and t�2. Then, the median of the market capitalizations as well as the 30 and 70% percentiles of the average stock returns were calculated leading to six portfolios based on these three values. The firms were allocated anew in each month t�1 over time to one of these six portfolios. Their construction only incorporates those corporations with avail- able market values for this and the subsequent month and additionally with available stock returns for the subsequent month t and for each month between t�12 and t�2. The resulting times-series of the value-weighted returns of four stock portfolios (between February 1997 and December 2006) with respect to the bottom and top 30% of the past average returns were the basis of the final calculations of MOMt which is the (weighted) difference between the returns of ‘winners’ and ‘losers’ in the past (according to the procedure described on the homepage of Kenneth R. French).

Table 2 reports descriptive statistics (mean, standard devi- ation, minimum and maximum) for the average monthly stock returns between 2003 and 2006 as well as for the esti- mated corporate beta parameters from the times-series regressions of the different asset pricing models. It addition- ally reports descriptive statistics for two important firm characteristics, namely the market value (in million US $) and the book-to-market equity at the end of the observation period 2006, respectively. Due to unavailable data and the exclusion of negative values of the book-to-market equity in Europe, the number of observations is only N = 170 for the market value and the book-to-market equity in the USA as well as only N = 253 for the market value and N = 247 for the book-to-market equity in Europe. Due to some extreme values, the median instead of the mean is used for these variables. The table shows, for example, that the mean average stock return for European corporations (2.34%) is noticeably higher than the mean for US corporations (1.28%) in this specific period.

982 U. von Arx and A. Ziegler

T ab le

1 . T h is

ta b le

re p o rt s ab so lu te

fr eq u en ci es

an d

re la ti v e fr eq u en ci es

(i n %

re g ar d in g

th e co lu m n s)

fo r th e v al u es

o f th e o rd in al

v ar ia b le s C o rp

i an d

In d u i in

2 0 0 2 . C o rp

i re fe rs

to th e

en v ir o n m en ta l an d so ci al

ac ti v it ie s o f fi rm

i co m p ar ed

w it h o th er

fi rm

s in

th e sa m e in d u st ry . In d u i sy m b o li ze s th e v ar ia b le

fo r en v ir o n m en ta l an d so ci al

p er fo rm

an ce

o f th e in d u st ry

to w h ic h fi rm

i b el o n g s. T h e n u m b er

o n e d es ig n at es

th e w o rs t an d th e n u m b er

fi v e th e b es t as se ss m en t fo r C o rp

i an d In d u i, re sp ec ti v el y.

In d u i

C o rp

i 5

4 3

2 1

T o ta l

U S A 5

1 (1 2 .5 0 % )

0 (0 .0 0 % )

6 (7 .2 3 % )

1 (2 .1 7 % )

0 (0 .0 0 % )

8 (4 .6 2 % )

4 0 (0 .0 0 % )

1 (3 .8 5 % )

1 0 (1 2 .0 5 % )

4 (8 .7 0 % )

1 (1 0 .0 0 % )

1 6 (9 .2 5 % )

3 4 (5 0 .0 0 % )

7 (2 6 .9 2 % )

2 6 (3 1 .3 3 % )

2 2 (4 7 .8 3 % )

6 (6 0 .0 0 % )

6 5 (3 7 .5 7 % )

2 3 (3 7 .5 0 % )

11 (4 2 .3 1 % )

3 7 (4 4 .5 8 % )

1 6 (3 4 .7 8 % )

2 (2 0 .0 0 % )

6 9 (3 9 .8 8 % )

1 0 (0 .0 0 % )

7 (2 6 .9 2 % )

4 (4 .8 2 % )

3 (6 .5 2 % )

1 (1 0 .0 0 % )

1 5 (8 .6 7 % )

T o ta l

8 (1 0 0 .0 0 % )

2 6 (1 0 0 .0 0 % )

8 3 (1 0 0 .0 0 % )

4 6 (1 0 0 .0 0 % )

1 0 (1 0 0 .0 0 % )

1 7 3 (1 0 0 .0 0 % )

In d u i

C o rp

i 5

4 3

2 1

T o ta l

E u ro p e

5 0 (0 .0 0 % )

2 (4 .1 7 % )

6 (5 .5 6 % )

4 (4 .8 8 % )

2 (8 .7 0 % )

1 4 (5 .2 2 % )

4 2 (2 8 .5 7 % )

1 0 (2 0 .8 3 % )

1 8 (1 6 .6 7 % )

2 3 (2 8 .0 5 % )

7 (3 0 .4 3 % )

6 0 (2 2 .3 9 % )

3 4 (5 7 .1 4 % )

2 2 (4 5 .8 3 % )

4 7 (4 3 .5 2 % )

4 0 (4 8 .7 8 % )

1 0 (4 3 .4 8 % )

1 2 3 (4 5 .9 0 % )

2 1 (1 4 .2 9 % )

1 0 (2 0 .8 3 % )

2 9 (2 6 .8 5 % )

1 4 (1 7 .0 7 % )

4 (1 7 .3 9 % )

5 8 (2 1 .6 4 % )

1 0 (0 .0 0 % )

4 (8 .3 3 % )

8 (7 .4 1 % )

1 (1 .2 2 % )

0 (0 .0 0 % )

1 3 (4 .8 5 % )

T o ta l

7 (1 0 0 .0 0 % )

4 8 (1 0 0 .0 0 % )

1 0 8 (1 0 0 .0 0 % )

8 2 (1 0 0 .0 0 % )

2 3 (1 0 0 .0 0 % )

2 6 8 (1 0 0 .0 0 % )

The effect of corporate social responsibility 983

6. Results

Table 3 reports the mutual correlation coefficients of �ri, the market value, the book-to-market equity and the CSR variables (including industry environmental and social per- formance). In this respect, Spearman’s rank correlation coefficients instead of Pearson’s correlation coefficients were applied when the ordinal variables Corpi and Indui or the market value and book-to-market equity are concerned. The main results in this table are the positive coefficients between stock performance and the different corporate CSR variables as well as the negative coefficients between the average stock returns and the industry environmental and social performance variables. Concerning the latter, they are strongly different from zero at least at the 5% sig- nificance level for the USA. Furthermore, the correlation

coefficients between the average stock returns and Corp543i for the USA and between stock performance and Corp54i for Europe are different from zero at the 5% significance level.

Table 4 reports the mutual Pearson’s correlation coeffi- cients of �ri and the estimated corporate beta parameters from the time-series regressions of the several asset pricing models. The main results in this table are the positive corre- lation coefficients between the average stock returns and the different estimated beta parameters. The only exception is the negative correlation coefficient between stock

performance and b̂ CAR

i4 for the USA, which is in addition different from zero at the 5% significance level. In contrast, the correlation coefficients between the average stock returns and the other estimated beta parameters in this

Table 2. This table reports several descriptive statistics for the average monthly stock returns �ri (in%) of firm i from 2003 to 2006 and the estimated beta parameters from the time-series regressions. In addition, the market value MVi (in million US $) and the book- to-market equity BMEi at the end of 2006 are considered. The numbers of observations for the USA are N = 170 for MVi and BMEi as well as N = 173 for the other variables. For Europe, the numbers are N = 253 for MVi, N = 247 for BMEi and N = 268 for the other variables. The smaller numbers of observations for MVi and BMEi are due to unavailable data and the exclusion of negative values of BMEi in Europe. For the case of MVi and BMEi, the median instead of the mean is used due to some extreme

values.

Variable Mean/median Standard deviation Minimum Maximum

USA �ri 1.28 0.93 �1.96 5.28 MVi 22134.63 67038.37 2400.04 455561.80

BMEi 0.30 0.17 0.06 0.78

b̂ CAPM

i 0.92 0.60 �0.30 3.00 b̂ FF

i1 1.08 0.50 �0.07 3.03 b̂ FF

i2 �0.06 0.41 �1.04 1.40 b̂ FF

i3 0.33 0.68 �1.78 2.37 b̂ CAR

i1 1.04 0.45 �0.09 2.87 b̂ CAR

i2 �0.04 0.42 �1.03 1.49 b̂ CAR

i3 0.31 0.68 �1.74 2.29 b̂ CAR

i4 �0.09 0.24 �1.13 0.48

Europe �ri 2.34 1.21 �8.84 6.95 MVi 13345.10 37158.86 997.98 219363.90

BMEi 0.36 0.21 0.01 1.17

b̂ CAPM

i 1.02 0.47 0.13 2.79

b̂ FF

i1 1.04 0.48 0.10 2.77

b̂ FF

i2 0.49 0.61 �1.82 2.77 b̂ FF

i3 0.16 0.61 �2.26 1.77 b̂ CAR

i1 1.01 0.44 0.18 2.59

b̂ CAR

i2 0.47 0.60 �1.85 2.75 b̂ CAR

i3 0.15 0.61 �2.26 1.76 b̂ CAR

i4 �0.09 0.25 �1.17 0.51

984 U. von Arx and A. Ziegler

T ab le

3 . T h is ta b le

re p o rt s th e m u tu al

co rr el at io n co ef fi ci en ts b et w ee n th e av er ag e m o n th ly

st o ck

re tu rn s � r i

(i n % ) o f fi rm

i fr o m

2 0 0 3 to

2 0 0 6 as

w el l as

th e m ar k et

v al u e M V i (i n m il li o n U S

$ ) an d th e b o o k -t o -m

ar k et

eq u it y B M E i at

th e en d o f 2 0 0 6 , re sp ec ti v el y.

F u rt h er m o re , th e o rd in al

v ar ia b le s C o rp

i an d In d u i in

2 0 0 2 ar e co n si d er ed

w h ic h re fe r to

th e en v ir o n m en ta l an d

so ci al

ac ti v it ie s o f fi rm

i co m p ar ed

w it h o th er

fi rm

s in

th e sa m e in d u st ry

as w el l as

to th e en v ir o n m en ta l an d so ci al

p er fo rm

an ce

o f th e in d u st ry

to w h ic h fi rm

i b el o n g s.

T h e n u m b er

o n e d es ig n at es

th e w o rs t an d th e n u m b er

fi v e th e b es t as se ss m en t fo r C o rp

i an d In d u i, re sp ec ti v el y.

In ad d it io n , th e d u m m y v ar ia b le s C o rp 5 4 i an d C o rp 5 4 3 i o n th e b as is o f C o rp

i as

w el l

as th e d u m m y v ar ia b le s In d u 5 4 i an d In d u 5 4 3 i o n th e b as is o f In d u i ar e ex am

in ed . If C o rp

i, In d u i, M V i, o r B M E i ar e co n si d er ed , S p ea rm

an ’s ra n k co rr el at io n s co ef fi ci en ts in st ea d o f P ea r-

so n ’s

co rr el at io n s co ef fi ci en ts

ar e u se d . T h e n u m b er s o f o b se rv at io n s fo r th e U S A

ar e N = 1 7 0 if M V i an d B M E i ar e co n si d er ed

as w el l as

N = 1 7 3 if o n ly

th e o th er

v ar ia b le s ar e co n si d -

er ed . F o r E u ro p e,

th e n u m b er s ar e N = 2 5 3

if M V i is

co n si d er ed , N = 2 4 7

if B M E i is

in cl u d ed

an d

N = 2 6 8

if o n ly

th e o th er

v ar ia b le s ar e co n si d er ed . T h e sm

al le r n u m b er s o f

o b se rv at io n s fo r M V i an d B M E i ar e d u e to

u n av ai la b le

d at a an d th e ex cl u si o n o f n eg at iv e v al u es

o f B M E i in

E u ro p e.

⁄ (⁄ ⁄ ,

⁄⁄ ⁄ )

m ea n s th at

th e co rr el at io n co ef fi ci en t is

d if fe re n t fr o m

ze ro

at th e 1 0 %

(5 an d 1 % ) si g n ifi ca n ce

le v el , re sp ec ti v el y.

� r i M V i

B M E i

C o rp

i In d u i

C o rp 5 4 i

In d u 5 4 i

C o rp 5 4 3 i

In d u 5 4 3 i

U S A � r i

1 .0 0

M V i

�0 .0 2

1 .0 0

B M E i

�0 .0 0

�0 .0 4

1 .0 0

C o rp

i 0 .1 2

0 .0 6

�0 .1 6 ⁄⁄

1 .0 0

In d u i

�0 .1 7 ⁄⁄

�0 .0 2

0 .0 1

�0 .1 4 ⁄

1 .0 0

C o rp 5 4 i

0 .0 5

0 .0 6

�0 .0 8

0 .6 4 ⁄⁄

⁄ �0

.0 1

1 .0 0

In d u 5 4 i

�0 .2 0 ⁄⁄

⁄ �0

.0 0

0 .0 9

�0 .1 8 ⁄⁄

0 .7 4 ⁄⁄

⁄ �0

.1 1

1 .0 0

C o rp 5 4 3 i

0 .1 6 ⁄⁄

0 .0 6

�0 .1 5 ⁄

0 .9 2 ⁄⁄

⁄ �0

.1 5 ⁄⁄

0 .3 9 ⁄⁄

⁄ �0

.1 3 ⁄

1 .0 0

In d u 5 4 3 i

�0 .1 7 ⁄⁄

�0 .0 2

�0 .0 1

�0 .0 8

0 .8 7 ⁄⁄

⁄ 0 .0 6

0 .3 4 ⁄⁄

⁄ �0

.1 3 ⁄

1 .0 0

E u ro p e

� r i 1 .0 0

M V i

�0 .1 5 ⁄⁄

1 .0 0

B M E i

�0 .1 2 ⁄

0 .0 8

1 .0 0

C o rp

i 0 .1 2 ⁄

0 .0 7

�0 .0 1

1 .0 0

In d u i

�0 .1 1 ⁄

�0 .1 5 ⁄⁄

�0 .0 5

�0 .1 4 ⁄⁄

1 .0 0

C o rp 5 4 i

0 .1 6 ⁄⁄

0 .0 4

�0 .0 4

0 .8 2 ⁄⁄

⁄ �0

.1 0 ⁄

1 .0 0

In d u 5 4 i

�0 .0 7

�0 .1 0

�0 .0 1

�0 .0 3

0 .7 4 ⁄⁄

⁄ �0

.0 2

1 .0 0

C o rp 5 4 3 i

0 .0 2

0 .0 8

0 .0 0

0 .8 1 ⁄⁄

⁄ �0

.1 1 ⁄

0 .3 7 ⁄⁄

⁄ �0

.0 1

1 .0 0

In d u 5 4 3 i

�0 .1 1 ⁄

�0 .1 0

�0 .0 5

�0 .1 7 ⁄⁄

⁄ 0 .8 9 ⁄⁄

⁄ �0

.1 2 ⁄

0 .4 1 ⁄⁄

⁄ �0

.1 5 ⁄⁄

1 .0 0

The effect of corporate social responsibility 985

country are clearly positive and different from zero at the 1% significance level. Regarding Europe, the respective cor- relation coefficients are without exception positive and only

insignificantly different from zero for b̂ CAPM

i and b̂ FF

i1 . However, it should be noted that the results in tables 3

and 4 only indicate univariate relationships. Therefore, tables 5 and 6 report the estimation results of our econo- metric analysis. The corresponding econometric models incorporate—besides the mainly interesting CSR variables (including industry environmental and social perfor- mance)—the estimated beta parameters from the time-series regressions of the several asset pricing models as financial control variables to explain the average monthly stock returns between 2003 and 2006. While table 5 refers to the USA, table 6 reports the estimation results for Europe. In both cases, the cross-sectional regressions according to equations (4), (4)′ and (4)″ are based on the CAPM, accord- ing to equations (5), (5)′ and (5)″ are based on the Fama- French three-factor model and according to equations (6), (6)′ and (6)″ are based on the Carhart four-factor model. Furthermore, the respective final regressions according to the first three equations (4), (5) and (6) incorporate the

ordinal variables Corpi and Indui, while the remaining regressions either include the dummies Corp54i and Indu54i or the dummies Corp543i and Indu543i.

According to table 5, the ordinal variable Corpi has a weakly positive effect (at the 10% significance level) when the estimated corporate beta parameters from the multifactor models are included as control variables. This impact is more significantly positive for the dummy variable Corp543i. In contrast, the latter effect has a higher signifi- cance level on the basis of the CAPM and the parameter of Corpi is not even significantly different from zero in this case. However, it appears that the estimation results based on the CAPM are less reliable because the estimated beta

parameters b̂ FF

i2 and b̂ CAR

i2 from the multifactor models—

besides b̂ CAPM

i , b̂ FF

i1 or b̂ CAR

i1 —have a high explanatory

power. Therefore, corporate environmental and social activi- ties obviously matter for the average monthly stock returns between 2003 and 2006 in the USA even when the effect is insignificant for the dummy Corp54i. This latter result points to possible non-linear effects with respect to the intensity of these measures.

Table 4. This table reports the mutual Pearson’s correlation coefficients between the average monthly stock returns �ri (in%) of firm i from 2003 to 2006 and the estimated beta parameters from the time-series regressions. The numbers of observations are N = 173 for the USA and N = 268 for Europe. ⁄ (⁄⁄, ⁄⁄⁄) means that the correlation coefficient is different from zero at the 10% (5 and

1%) significance level, respectively.

�ri b̂ CAPM

i b̂ FF

i k̂ FF

i1 k̂ FF

i2 b̂ CAR

i k̂ CAR

i1 k̂ CAR

i2 k̂ CAR

i3

USA �ri 1.00

b̂ CAPM

i 0.27 ⁄⁄⁄ 1.00

b̂ FF

i1 0.38 ⁄⁄⁄ 0.86⁄⁄⁄ 1.00

b̂ FF

i2 0.47 ⁄⁄⁄ 0.32⁄⁄⁄ 0.38⁄⁄⁄ 1.00

b̂ FF

i3 0.21 ⁄⁄⁄ �0.44⁄⁄⁄ 0.07 0.28⁄⁄⁄ 1.00

b̂ CAR

i1 0.38 ⁄⁄⁄ 0.85⁄⁄⁄ 0.98⁄⁄⁄ 0.38⁄⁄⁄ 0.06 1.00

b̂ CAR

i2 0.47 ⁄⁄⁄ 0.37⁄⁄⁄ 0.44⁄⁄⁄ 0.99⁄⁄⁄ 0.28⁄⁄⁄ 0.41⁄⁄⁄ 1.00

b̂ CAR

i3 0.20 ⁄⁄⁄ �0.47⁄⁄⁄ 0.03 0.27⁄⁄⁄ 1.00⁄⁄⁄ 0.03 0.26⁄⁄⁄ 1.00

b̂ CAR

i4 �0.17⁄⁄ �0.43⁄⁄⁄ �0.52⁄⁄⁄ �0.18⁄⁄ �0.05 �0.35⁄⁄⁄ �0.31⁄⁄⁄ 0.02 1.00

Europe

�ri 1.00

b̂ CAPM

i 0.05 1.00

b̂ FF

i1 0.08 1.00 ⁄⁄⁄ 1.00

b̂ FF

i2 0.35 ⁄⁄⁄ 0.17⁄⁄⁄ 0.22⁄⁄⁄ 1.00

b̂ FF

i3 0.16 ⁄⁄ �0.39⁄⁄⁄ �0.37⁄⁄⁄ 0.09 1.00

b̂ CAR

i1 0.12 ⁄⁄ 0.98⁄⁄⁄ 0.98⁄⁄⁄ 0.21⁄⁄⁄ �0.38⁄⁄⁄ 1.00

b̂ CAR

i2 0.36 ⁄⁄⁄ 0.11⁄ 0.17⁄⁄⁄ 0.99⁄⁄⁄ 0.10 0.17⁄⁄⁄ 1.00

b̂ CAR

i3 0.16 ⁄⁄ �0.40⁄⁄⁄ �0.37⁄⁄⁄ 0.09 1.00⁄⁄⁄ �0.38⁄⁄⁄ 0.10 1.00

b̂ CAR

i4 0.13 ⁄⁄ �0.56⁄⁄⁄ �0.56⁄⁄⁄ �0.18⁄⁄⁄ 0.11⁄ �0.40⁄⁄⁄ �0.08 0.13⁄⁄ 1.00

986 U. von Arx and A. Ziegler

T ab le

5 . T h is

ta b le

re p o rt s th e O L S p ar am

et er

es ti m at es

in th e fi n al

cr o ss -s ec ti o n al

re g re ss io n s fo r th e U S A . T h e d ep en d en t v ar ia b le

is th e av er ag e m o n th ly

st o ck

re tu rn

� r i (i n % ) o f fi rm

i fr o m

2 0 0 3 to

2 0 0 6 .T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 ), (5 ) an d (6 ) ar e th e o rd in al

v ar ia b le s C o rp

i an d In d u i in

2 0 0 2 w h ic h re fe r to

th e en v ir o n m en ta l an d so ci al

ac ti v it ie s o f fi rm

i co m -

p ar ed

w it h o th er

fi rm

s in

th e sa m e in d u st ry

as w el l as

to th e en v ir o n m en ta l an d so ci al

p er fo rm

an ce

o f th e in d u st ry

to w h ic h fi rm

i b el o n g s.

T h e n u m b er

o n e d es ig n at es

th e w o rs t an d

th e n u m b er

fi v e th e b es t as se ss m en t fo r C o rp

i an d In d u i, re sp ec ti v el y.

T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 )′ , (5 )′ an d (6 )′ ar e th e d u m m y v ar ia b le s C o rp 5 4 i o n th e b as is

o f C o rp

i an d

In d u 5 4 i o n th e b as is

o f In d u i. T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 )″ , (5 )″

an d (6 )″

ar e th e d u m m y v ar ia b le s C o rp 5 4 3 i o n th e b as is

o f C o rp

i an d In d u 5 4 3 i o n th e b as is

o f In d u i. W h il e

(4 ), (4 )′ an d (4 )″

ar e b as ed

o n th e C A P M

an d th u s o n ly

in co rp o ra te

b̂ C A P M

i as

co n tr o l v ar ia b le , (5 ), (5 )′ an d (5 )″

ar e b as ed

o n th e F am

a– F re n ch

th re e- fa ct o r m o d el

in cl u d in g b̂ F F

i1 , b̂ F F

i2

an d b̂ F F

i3 as

co n tr o l v ar ia b le s an d (6 ), (6 )′ an d (6 )″

ar e b as ed

o n th e C ar h ar t fo u r- fa ct o r m o d el

in cl u d in g b̂ C A R

i1 , b̂ C A R

i2 , b̂ C A R

i3 an d b̂ C A R

i4 as

co n tr o l v ar ia b le s.

T h e n u m b er s o f o b se rv at io n s

ar e N = 1 7 3 , re sp ec ti v el y.

⁄ (⁄ ⁄ ,

⁄⁄ ⁄ )

m ea n s th at

th e ap p ro p ri at e p ar am

et er

is d if fe re n t fr o m

ze ro

o r—

re g ar d in g th e F -t es t—

th at

al l ex p la n at o ry

v ar ia b le s to g et h er

h av e an

ef fe ct

at th e

1 0 %

(5 an d 1 % ) si g n ifi ca n ce

le v el , re sp ec ti v el y.

T h e st an d ar d d ev ia ti o n s fo r th e es ti m at ed

p ar am

et er s ar e es ti m at ed

h et er o sc ed as ti ci ty

ro b u st .

E x p la n at o ry

v ar ia b le s

(4 )

(5 )

(6 )

(4 )′

(5 )′

(6 )′

(4 )″

(5 )″

(6 )″

C o rp

i 0 .0 8

0 .0 9 ⁄

0 .1 0 ⁄

– –

– –

– –

In d u i

�0 .2 2 ⁄⁄

⁄ �0

.0 8

�0 .0 8

– –

– –

– –

C o rp 5 4 i

– –

– �0

.0 8

0 .0 3

0 .0 4

– –

–

In d u 5 4 i

– –

– �0

.4 4 ⁄⁄

⁄ �0

.2 5

�0 .2 5

– –

–

C o rp 5 4 3 i

– –

– –

– –

0 .2 2 ⁄

0 .2 3 ⁄⁄

0 .2 3 ⁄⁄

In d u 5 4 3 i

– –

– –

– –

�0 .5 0 ⁄⁄

⁄ �0

.2 5 ⁄

�0 .2 5 ⁄

b̂ C A P M

i 0 .4 7 ⁄⁄

⁄ –

– 0 .4 2 ⁄⁄

⁄ –

– 0 .5 3 ⁄⁄

⁄ –

b̂ F F

i1 –

0 .4 6 ⁄⁄

⁄ –

– 0 .4 4 ⁄⁄

⁄ –

– 0 .5 1 ⁄⁄

⁄ –

b̂ F F

i2 –

0 .7 3 ⁄⁄

⁄ –

– 0 .7 5 ⁄⁄

⁄ –

– 0 .7 1 ⁄⁄

⁄ –

b̂ F F

i3 –

0 .1 3

– –

0 .1 3

– –

0 .0 9

–

b̂ C A R

i1 –

– 0 .5 1 ⁄⁄

⁄ –

– 0 .4 7 ⁄⁄

⁄ –

– 0 .5 5 ⁄⁄

⁄

b̂ C A R

i2 –

– 0 .7 2 ⁄⁄

⁄ –

– 0 .7 4 ⁄⁄

⁄ –

– 0 .7 1 ⁄⁄

⁄

b̂ C A R

i3 –

– 0 .1 3

– –

0 .1 4

– –

0 .1 0

b̂ C A R

i4 –

– 0 .1 0

– –

0 .1 0

– –

0 .0 9

C o n st an t

1 .2 8 ⁄⁄

⁄ 0 .7 7 ⁄⁄

0 .7 3 ⁄⁄

0 .9 9 ⁄⁄

⁄ 0 .8 5 ⁄⁄

⁄ 0 .8 2 ⁄⁄

⁄ 1 .0 1 ⁄⁄

⁄ 0 .7 9 ⁄⁄

⁄ 0 .7 5 ⁄⁄

⁄

R 2

0 .1 3

0 .2 9

0 .2 9

0 .1 1

0 .2 9

0 .2 9

0 .1 5

0 .3 1

0 .3 1

F -v al u e

6 .1 6 ⁄⁄

⁄ 1 8 .1 0 ⁄⁄

⁄ 1 4 .8 7 ⁄⁄

⁄ 5 .0 2 ⁄⁄

⁄ 1 7 .1 2 ⁄⁄

⁄ 1 4 .0 5 ⁄⁄

⁄ 8 .2 3 ⁄⁄

⁄ 1 9 .1 2 ⁄⁄

⁄ 1 5 .7 3 ⁄⁄

⁄

The effect of corporate social responsibility 987

T ab le

6 . T h is ta b le

re p o rt s th e O L S p ar am

et er

es ti m at es

in th e fi n al

cr o ss -s ec ti o n al

re g re ss io n s fo r E u ro p e. T h e d ep en d en t v ar ia b le

is th e av er ag e m o n th ly

st o ck

re tu rn

� r i (i n % ) o f fi rm

i fr o m

2 0 0 3

to 2 0 0 6 .T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 ), (5 ) an d (6 ) ar e th e o rd in al

v ar ia b le s C o rp

i an d In d u i in

2 0 0 2 w h ic h re fe r to

th e en v ir o n m en ta l an d so ci al

ac ti v it ie s o f fi rm

i co m p ar ed

w it h

o th er

fi rm

s in

th e sa m e in d u st ry

as w el l as

to th e en v ir o n m en ta l an d so ci al

p er fo rm

an ce

o f th e in d u st ry

to w h ic h fi rm

i b el o n g s. T h e n u m b er

o n e d es ig n at es

th e w o rs t an d th e n u m b er

fi v e

th e b es t as se ss m en t fo r C o rp

i an d In d u i, re sp ec ti v el y.

T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 )′ , (5 )′ an d (6 )′ ar e th e d u m m y v ar ia b le s C o rp 5 4 i o n th e b as is o f C o rp

i an d In d u 5 4 i o n th e b as is

o f In d u i. T h e m ai n ex p la n at o ry

v ar ia b le s in

(4 )″ , (5 )″

an d (6 )″

ar e th e d u m m y v ar ia b le s C o rp 5 4 3 i o n th e b as is

o f C o rp

i an d In d u 5 4 3 i o n th e b as is

o f In d u i. W h il e (4 ), (4 )′ an d (4 )″

ar e

b as ed

o n th e C A P M

an d th u s in co rp o ra te

b̂ C A P M

i as

co n tr o l v ar ia b le , (5 ), (5 )′ an d (5 )″

ar e b as ed

o n th e F am

a– F re n ch

th re e- fa ct o r m o d el

in cl u d in g b̂ F F

i1 , b̂ F F

i2 an d b̂ F F

i3 as

co n tr o l v ar ia b le s

an d (6 ), (6 )′ an d (6 )″

ar e b as ed

o n th e C ar h ar t fo u r- fa ct o r m o d el

in cl u d in g b̂ C A R

i1 , b̂ C A R

i2 , b̂ C A R

i3 an d b̂ C A R

i4 as

co n tr o l v ar ia b le s. F u rt h er m o re , se v er al

co u n tr y d u m m ie s ar e in co rp o ra te d as

co n -

tr o l v ar ia b le s. T h e n u m b er s o f o b se rv at io n s ar e N = 2 6 8 , re sp ec ti v el y.

⁄ (⁄ , ⁄⁄

⁄ ) m ea n s th at

th e ap p ro p ri at e p ar am

et er

is d if fe re n t fr o m

ze ro

o r—

re g ar d in g th e F -t es t—

th at

al l ex p la n at o ry

v ar ia b le s to g et h er

h av e an

ef fe ct

at th e 1 0 %

(5 an d 1 % ) si g n ifi ca n ce

le v el , re sp ec ti v el y. T h e st an d ar d d ev ia ti o n s fo r th e es ti m at ed

p ar am

et er s ar e es ti m at ed

h et er o sc ed as ti ci ty

ro b u st .

E x p la n at o ry

v ar ia b le s

(4 )

(5 )

(6 )

(4 )′

(5 )′

(6 )′

(4 )″

(5 )″

(6 )″

C o rp

i 0 .0 9

0 .0 9

0 .0 8

– –

– –

– –

In d u i

�0 .1 2 ⁄

0 .0 2

0 .0 0

– –

– –

– –

C o rp 5 4 i

– –

– 0 .3 4 ⁄⁄

0 .3 1 ⁄⁄

0 .3 2 ⁄⁄

– –

–

In d u 5 4 i

– –

– �0

.2 3

0 .1 0

0 .0 7

– –

–

C o rp 5 4 3 i

– –

– –

– –

�0 .0 9

�0 .0 9

�0 .1 3

In d u 5 4 3 i

– –

– –

– –

�0 .3 6 ⁄⁄

�0 .1 8

�0 .2 1

b̂ C A P M

i 0 .2 5

– –

0 .2 3

– –

0 .2 6

– –

b̂ F F

i1 –

0 .1 9

– –

0 .1 9

– –

0 .2 5

–

b̂ F F

i2 –

0 .5 8 ⁄⁄

⁄ –

– 0 .5 8 ⁄⁄

⁄ –

– 0 .5 4 ⁄⁄

⁄ –

b̂ F F

i3 –

0 .3 0 ⁄⁄

– –

0 .3 2 ⁄

– –

0 .3 1 ⁄⁄

–

b̂ C A R

i1 –

– 0 .6 3 ⁄⁄

– –

0 .6 4 ⁄⁄

– –

0 .7 1 ⁄⁄

b̂ C A R

i2 –

– 0 .5 9 ⁄⁄

⁄ –

– 0 .6 0 ⁄⁄

⁄ –

– 0 .5 5 ⁄⁄

⁄

b̂ C A R

i3 –

– 0 .3 6 ⁄⁄

– –

0 .3 8 ⁄⁄

– –

0 .3 8 ⁄⁄

b̂ C A R

i4 –

– 0 .8 7

– –

0 .8 9

– –

0 .8 9

F in

i �0

.1 5

�0 .1 6

�0 .2 1

�0 .1 9

�0 .1 9

�0 .2 5

�0 .1 3

�0 .1 2

�0 .1 8

F ra

i �0

.9 1 ⁄⁄

⁄ �0

.7 0 ⁄⁄

⁄ �0

.5 4 ⁄⁄

�0 .8 6 ⁄⁄

⁄ �0

.6 5 ⁄⁄

⁄ �0

.4 8 ⁄

�0 .9 7 ⁄⁄

⁄ �0

.7 9 ⁄⁄

⁄ �0

.6 3 ⁄⁄

⁄

G er

i �0

.3 6

�0 .3 3

�0 .2 0

�0 .3 0

�0 .3 0

�0 .1 6

�0 .3 9

�0 .4 0

�0 .2 6

It a i

�0 .5 8 ⁄⁄

�0 .6 3 ⁄⁄

�0 .5 9 ⁄⁄

�0 .5 8 ⁄⁄

�0 .6 5 ⁄⁄

⁄ �0

.6 0 ⁄⁄

�0 .7 4 ⁄⁄

�0 .7 9 ⁄⁄

⁄ �0

.7 7 ⁄⁄

⁄

N et i

�1 .1 5 ⁄⁄

⁄ �0

.9 7 ⁄⁄

⁄ �0

.7 7 ⁄⁄

�1 .1 6 ⁄⁄

⁄ �0

.9 7 ⁄⁄

⁄ �0

.7 7 ⁄⁄

�1 .1 5 ⁄⁄

⁄ �0

.9 7 ⁄⁄

⁄ �0

.7 6 ⁄⁄

S p a i

�0 .3 5

�0 .1 7

�0 .2 3

�0 .3 1

�0 .1 6

�0 .2 1

�0 .4 9 ⁄

�0 .2 9

�0 .3 7

S w e i

�0 .2 9

�0 .3 2

�0 .0 3

�0 .4 0

�0 .3 6

�0 .0 9

�0 .1 7

�0 .2 2

0 .0 9

S w i i

�0 .3 6

�0 .4 7 ⁄

�0 .3 5

�0 .3 5

�0 .4 3

�0 .3 2

�0 .3 4

�0 .4 4

�0 .3 1

U K i

�0 .7 3 ⁄⁄

⁄ �0

.7 0 ⁄⁄

⁄ �0

.4 0 ⁄

�0 .7 0 ⁄⁄

⁄ �0

.6 8 ⁄⁄

⁄ �0

.3 7

�0 .7 5 ⁄⁄

⁄ �0

.7 1 ⁄⁄

⁄ �0

.4 0 ⁄

C o n st an t

2 .6 9 ⁄⁄

⁄ 1 .9 9 ⁄⁄

⁄ 1 .5 5 ⁄⁄

⁄ 2 .5 9 ⁄⁄

⁄ 2 .1 9 ⁄⁄

⁄ 1 .6 6 ⁄⁄

⁄ 2 .9 2 ⁄⁄

⁄ 2 .4 7 ⁄⁄

⁄ 1 .9 7 ⁄⁄

⁄

R 2

0 .0 9

0 .2 0

0 .2 3

0 .1 0

0 .2 1

0 .2 4

0 .1 0

0 .2 0

0 .2 4

F -v al u e

2 .6 3 ⁄⁄

⁄ 3 .7 6 ⁄⁄

⁄ 4 .0 2 ⁄⁄

⁄ 2 .8 8 ⁄⁄

⁄ 3 .9 3 ⁄⁄

⁄ 4 .3 6 ⁄⁄

⁄ 2 .5 4 ⁄⁄

⁄ 3 .7 6 ⁄⁄

⁄ 4 .0 4 ⁄⁄

⁄

988 U. von Arx and A. Ziegler

Concerning the impact of industry environmental and social performance for the US stock market, the estimation results are different. The ordinal variable Indui as well as the corresponding dummy variables Indu54i and Indu543i have a strong negative effect on stock performance (at the 1% significance level) when the estimated beta parameters from the CAPM are incorporated. However, these effects become much less significant for the case of Indu543i on the basis of the Fama–French three-factor or the Carhart four-factor models (the corresponding slightly significant effect would be in line with some results in Hong and Kacperczyk 2009, who report in their analysis of sin stocks in the USA that a lower industry social performance has a positive effect on stock performance). Furthermore, the effects become even completely insignificant for the two other industry environmental and social performance vari- ables on the basis of the multifactor models. Due to the high explanatory power of some estimated corporate beta parameters from the multifactor models as discussed above, the estimation results based on the CAPM are overall obvi- ously not reliable, but a typical example for biased parame- ter estimations due to omitted explanatory variables.

According to table 6, this problem holds true for the effect of Indu543i and (to a lower extent) of Indui in Eur- ope. While these variables have a negative influence on average stock returns (at the 5 or 10% significance level) based on the CAPM, this effect becomes insignificant on the basis of both multifactor models. The impacts of Indu54i are already insignificant, irrespective of the underly- ing asset pricing models. In the same way, no significant effects of Corpi and Corp543i arise. In contrast, Corp54i has a positive effect on stock performance (at the 5% sig- nificance level) based on the CAPM as well as on both multifactor models. Therefore, only single positive effects on average monthly stock returns between 2003 and 2006 appear to be existent in Europe, but no linear effects of an increasing intensity of corporate environmental and social activities.

These estimation results are robust in different ways: in order to account for our two-stage econometric analysis which includes estimated beta parameters from the first stage in the final cross-sectional regressions, we have also con- ducted adjustments of the estimated standard deviations of the estimated parameters according to Shanken (1992). While the estimated beta parameters now—as expected—mostly have no significant effect on stock performance (an exception is that in Europe b̂

FF

i2 still has a significantly positive impact), the estimation results for the mainly interesting CSR vari- ables are not systematically different from the results as dis- cussed above. In fact, the adjusted estimation results (which are available on request) even strengthen some previous con- clusions since Indu543i, for example, in the USA now has a completely insignificant effect on the basis of the Fama– French three-factor and the Carhart four-factor models.

Furthermore, it should be noted that corporations were ranked each year on their market capitalization in June and on their book-to-market equity in December of the previous year for the calculation of SMBt and HMLt in Europe as discussed above. In this respect, we used the book values which were published in June each year in the Thomson

Financial Datastream database to ensure that the values for the previous year are actually considered. However, it is also possible that these values are influenced by new devel- opments during the current year. Therefore, we additionally considered the book-to-market equity based on the pub- lished book and market values in December of the previous year in a further analysis. Moreover, we also examined the published book and market values in June from Thomson Financial Datastream. Indeed, the estimation results based on these calculations are not systematically different from the main results as discussed above.

Finally, the estimation results in this paper are in princi- ple based on the 2002 assessments of CSR (including industry environmental and social performance). However, it should be noted that many corporations were assessed for the first time by Sarasin after 2002. For these firms, we incorporated the corresponding first assessments in our empirical analysis. This procedure seems to be justified because the assessments have an extremely low variability over time for the respective corporations. We nevertheless excluded in a further analysis those firms with very recent assessments, i.e. with first assessments after 2004. However, we continued to include firms with first assessments in 2003 or 2004. This procedure of extending assessments for two years is common in empirical analyses of the relation- ship between CSR and corporate financial performance to avoid very small samples (e.g. Derwall et al. 2005). Indeed, the corresponding estimation results are again qualitatively fully in line with the main results as discussed above.

7. Discussion and conclusions

This paper provides new empirical evidence for the effect of CSR (which is measured by environmental and social activities of a firm compared with other firms within the industry and additionally considers environmental and social performance of the industry to which a firm belongs) on average monthly stock returns between 2003 and 2006. In contrast to former studies, it examines two worldwide leading stock markets, namely the USA and the European stock markets, in order to analyse potentially different rela- tionships between CSR and financial performance. Our two-stage econometric analysis shows that corporate envi- ronmental and social activities matter for the explanation of stock performance in both regions. However, this impact is obviously not linear for an increasing intensity of these measures. Compared with Europe, the positive effect fur- thermore is more robust for the USA. While it can only be speculated why the positive impact in the USA is slightly stronger, one explanation could be the longer tradition of ethical components of CSR and particularly of SRI than in Europe. In contrast, the industry environmental and social performance has no robust influence on the average monthly stock returns between 2003 and 2006 in any region.

According to these results, the stock markets—and partic- ularly the US stock market—obviously rewarded invest- ments in stocks of corporations with a high intensity of

The effect of corporate social responsibility 989

environmental and social activities compared with other firms within the industry. In other words, investors who applied a buy-and-hold strategy would have increased their portfolio value by investing in such stocks. Regarding the management of a firm, these results imply that such mea- sures could be increased since they obviously do not lead to worse financial performance. The results furthermore support the advocates of information-based regulations by improving the flow of the respective information. However, the question is whether the discussed positive effect is robust for alternative measurements of CSR, for example, based on assessments from other rating agencies or based on quantitative and thus more objective indicators such as emissions, as well as for alternative measurements of corpo- rate financial performance, for example, on the basis of accounting data based indicators. Such studies would be interesting in the future. Another field for further research would be the econometric analysis of alternative periods to examine whether the consideration of the period between 2003 and 2006 produces specific estimation results.

Irrespective of such future research, our study supports the incorporation of more flexible asset pricing models: on the basis of the simple CAPM, industry environmental and social performance has a significantly negative impact on stock performance in the USA. However, the significance of this effect strongly decreases and mostly disappears if estimated corporate beta parameters from the Fama–French three-factor or the Carhart four-factor models are included as additional control variables. This result (in line with e.g. McWilliams and Siegel 2000) points to the problem of mis- leading conclusions regarding the effect of CSR on corpo- rate financial performance if misspecified econometric models are applied due to omitted explanatory variables such that biased parameter estimations occur.

Acknowledgements

We would like to thank a referee for his useful comments, Eckhard Plinke and the bank Sarasin & Cie in Basle for providing their assessment data, Kenneth R. French, Ulrich Oberndorfer, Michael Schröder and participants of several conferences for stimulating discussions, as well as Eveline Schwegler for her untiring commitment during data analysis. Our special thanks go to Peter Schmidt for his very helpful support in the econometric analysis.

References

Alberini, A. and Segerson, K., Assessing voluntary programs to improve environmental quality. Environ. Resour. Econ., 2002, 22, 157–184.

Banz, R.W., The relationship between return and market value of common stocks. J. Financ. Econ., 1981, 9, 3–18.

Barnett, M.L. and Salomon, R.M., Beyond dichotomy: The curvi- linear relationship between social responsibility and financial performance. Strateg. Manag. J., 2006, 27, 1101–1122.

Barney, J., Firm resources and sustained competitive advantage. J. Manage., 1991, 17, 99–120.

Bauer, R., Derwall, J. and Otten, R., The ethical mutual fund per- formance debate: New evidence from Canada. J. Bus. Ethics, 2007, 70, 111–124.

Bauer, R., Koedijk, K. and Otten, R., International evidence on ethical mutual fund performance and investment style. J. Bank. Finance, 2005, 29, 1751–1767.

Becchetti, L., Di Giacomo, S. and Pinnacchio, D., Corporate social responsibility and corporate performance: Evidence from a panel of US listed companies. Appl. Econ., 2008, 40, 541–567.

Beltratti, A., Capital market equilibrium with externalities, produc- tion and heterogeneous agents. J. Bank. Finance, 2005, 29, 3061–3073.

Berkowitz, M.K. and Qiu, J., Common risk factors in explaining Canadian equity returns. Working paper, University of Toronto, 2001.

Bollen, N.P.B. and Busse, J.A., Short-term persistence in mutual fund performance. Rev. Financ. Stud., 2005, 18, 569–597.

Carhart, M.M., On persistence in mutual fund performance. J. Financ., 1997, 52, 57–82.

Curran, M.M. and Moran, D., Impact of the FTSE4Good index on firm price: An event study. J. Environ. Manage., 2007, 82, 529– 537.

Dasgupta, S., Laplante, B. and Nlandu, M., Pollution and capital markets in developing countries. J. Environ. Econ. Manage., 2001, 42, 310–335.

DeBondt, W.F.M. and Thaler, R., Does the stock market overreact? J. Financ., 1985, 40, 793–805.

Derwall, J., Guenster, N., Bauer, R. and Koedijk, K., The eco-effi- ciency premium puzzle. Financ. Analyst. J., 2005, 61, 51–63.

Elsayed, K. and Paton, D., The impact of environmental perfor- mance on firm performance: Static and dynamic panel data evi- dence. Struct. Change Econ. Dynamics, 2005, 16, 395–412.

Fama, E.F. and French, K.R., The cross-section of expected stock returns. J. Financ., 1992, 47, 427–465.

Fama, E.F. and French, K.R., Common risk factors in the returns on stocks and bonds. J. Financ. Econ., 1993, 33, 3–56.

Fama, E.F. and French, K.R., Multifactor explanations of asset pricing anomalies. J. Financ., 1996, 51, 55–84.

Fama, E.F. and French, K.R., The capital asset pricing model: Theory and evidence. J. Econ. Perspect., 2004, 18, 25–46.

Filbeck, G. and Gorman, R.F., The relationship between the envi- ronmental and financial performance of public utilities. Environ. Resour. Econ., 2004, 29, 137–157.

Friedman, M., The social responsibility of business is to increase its profits. The New York Times Magazine, 13 September, 1970.

Guenster, N., Bauer, R., Derwall, J. and Koedijk, K., The eco- nomic value of corporate eco-efficiency. Eur. Financ. Manage., 2011, 17, 679–704.

Gupta, S. and Goldar, B., Do stock markets penalize environment- unfriendly behaviour? Evidence from India Ecol. Econ., 2005, 52, 81–95.

Hamilton, J.T., Pollution as news: Media and stock market reac- tions to the toxics release inventory data. J. Environ. Econ. Manage., 1995, 28, 98–113.

Hart, S.L. and Ahuja, G., Does it pay to be green? An empirical examination of the relationship between emission reduction and firm performance Business Strat. Environ., 1996, 5, 30–37.

Heal, G., Corporate social responsibility: An economic and finan- cial framework. Geneva Papers, 2005, 30, 387–409.

Heinkel, R., Kraus, A. and Zechner, J., The effect of green invest- ment on corporate behavior. J. Financ. Quantitat. Anal., 2001, 36, 431–449.

Hong, H. and Kacperczyk, M., The price of sin: The effects of social norms on markets. J. Financ. Econ., 2009, 93, 15–36.

Hussain, I., Toms, S. and Diacon, S., Financial distress, single and multifactor tests and comparisons of asset pricing anomalies: new evidence. Working Paper, Nottingham University Business School, 2002.

Jagadeesh, N. and Titman, S., Returns to buying winners and sell- ing losers: Implications from stock market efficiency. J. Financ., 1993, 48, 65–91.

990 U. von Arx and A. Ziegler

Jagadeesh, N. and Titman, S., Profitability of momentum strate- gies: An evaluation of alternative explanations. J. Financ., 2001, 56, 699–720.

Kempf, A. and Osthoff, P., The effect of socially responsible investing on portfolio performance. Eur. Financ. Manage., 2007, 13, 908–922.

King, A. and Lenox, M., Does it really pay to be green? J. Ind. Ecol., 2001, 5, 105–116.

King, A. and Lenox, M., Exploring the locus of profitable pollu- tion reduction. Manage. Sci., 2002, 48, 289–299.

Klassen, R.D. and Whybark, D.C., The impact of environmental technologies on manufacturing performance. Acad. Manag. J., 1999, 42, 599–615.

Kothari, S.P. and Warner, J.B., Econometrics of event studies. In Handbook of Corporate Finance: Empirical Corporate Finance, edited by B.E. Eckbo, Chapter 1, pp. 3–36, 2006 (Elsevier: North Holland).

L’Her, J.-F., Masmoudi, T. and Suret, J.-M., Evidence to sup- port the four-factor pricing model from the Canadian stock market. J. Int. Financ. Markets Institut. Money, 2004, 14, 313–328.

Lintner, J., The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Rev. Econ. Stat., 1965, 47, 13–37.

MacKinlay, A.C., Event studies in economics and finance. J. Econ. Literat., 1997, 35, 13–39.

McWilliams, A. and Siegel, D., Corporate social responsibility and financial performance: Correlation or misspecification? Strateg. Manag. J., 2000, 21, 603–609.

McWilliams, A. and Siegel, D., Corporate social responsibility: A theory of the firm perspective. Acad. Manag. Rev., 2001, 26, 117–127.

McWilliams, A., Siegel, D. and Wright, P.M., Corporate social respon- sibility: Strategic implications. J. Manage. Stud., 2006, 43, 1–18.

Miller, M. and Modigliani, F., Dividend policy, growth and the valuation of shares. J. Bus., 1961, 34, 411–433.

Orlitzky, M., Does firm size confound the relationship between corporate social performance and firm performance. J. Bus. Eth- ics, 2001, 33, 167–180.

Orlitzky, M., Schmidt, F.L. and Rynes, S.L., Corporate social and financial performance: A meta-analysis. Organizat. Stud., 2003, 24, 403–441.

Perold, A.F., The capital asset pricing model. J. Econ. Perspect., 2004, 18, 3–24.

Posnikoff, J.F., Disinvestment from South Africa: They did well by doing good. Contemp. Econ. Policy, 1997, 15, 76–86.

Rouwenhorst, K.G., International momentum strategies. J. Financ., 1998, 53, 267–284.

Russo, M.V. and Fouts, P.A., A resource-based perspective on cor- porate environmental performance and profitability. Acad. Manag. J., 1997, 40, 534–559.

Shanken, J., On the estimation of beta-pricing models. Rev. Financ. Stud., 1992, 5, 1–33.

Sharpe, W.F., A simplified model for portfolio analysis. Manage. Sci., 1963, 9, 277–293.

Shleifer, A. and Vishny, R.W., A survey of corporate governance. J. Financ., 1997, 52, 737–783.

Telle, K., ‘It pays to be green’—a premature conclusion? Environ. Resour. Econ., 2006, 35, 195–220.

Tirole, J., The Theory of Corporate Finance, 2006 (Princeton Uni- versity Press: Princeton, NJ).

Waddock, S. and Graves, S.B., The corporate social perfor- mance—financial performance link. Strateg. Manag. J., 1997, 18, 303–319.

Ziegler, A., Is it beneficial to be included in a sustainability stock index: A panel data study for European firms. Environ. Resource Econ., 2012, 52, 301–325.

Ziegler, A., Busch, T. and Hoffmann, V.H., Disclosed corporate responses to climate change and stock performance: An interna- tional empirical analysis. Energy Econ., 2011, 33, 1283–1294.

Ziegler, A., Schröder, M. and Rennings, K., The effect of environ- mental and social performance on the stock performance of Euro- pean corporations. Environ. Resour. Econ., 2007a, 37, 661–680.

Ziegler, A., Schröder, M., Schulz, A. and Stehle, R., Multifaktormo- delle zur Erklärung deutscher Aktienrenditen: Eine empirische Analyse [Multifactor models for the explanation of German stock returns: An empirical analysis]. Schmalenbachs Zeitschrift für betriebswirtschaftliche Forschung, 2007b, 59, 355–389.

The effect of corporate social responsibility 991

Copyright of Quantitative Finance is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.